460 1gb and 768mb

[robgmun]

Overclock the 460 768 and that 9 fps difference is no more.

But then you can clock the 460 1Gb to the same level and the difference is back (and maybe more)

Wayne - there's one big thing you are forgetting. Future proofing, as games use more VRAM the 768 will begin to struggle more and more and the 1GB will pull further ahead and the investment will be clear, 12 months from now the difference in fps between the two will be greater as the 768 will drop to single figures. It's called investment

 

[arknor]

most games already use more than 768mb of memory at 1080.

as i said min/max are useless when your stuttering along out of memory

 

[NqR]

most games already use more than 768mb of memory at 1080.

as i said min/max are useless when your stuttering along out of memory

min in benchmarks isn't suppose to include stuttering?

I mean if it stutters it will have a negative impact on min fps

 

[MikeB]

the vram argument was in full swing 4 years ago when i spent close on £400 on a 8800 gtx, far as I know the memory on that was never used fully in its lifetime.

Future proofing and what if scenarios on PC's don't work, most PC games are ports nowdays and they don't come close to stressing the cards unless very badly written

 

[robgmun]

Look at Arknor poster again then, because he's right, for gaming in the next 24months, you need more then 768

 

[TheRealDeal]

Overclock the 460 768 and that 9 fps difference is no more.

The overclock arguement is stupid tbh as both the gtx460 and 5850 overclock every bit as good as the gtx460 768mb.

The gtx768 will do a fine job but for piece of mind the extra rops and memory that comes with the gtx460 1gb is where my money would be going.

 

[RavenXXX2]

lol, we're talking about saving money and overlocking to get performance past the 460 and even 5850, the fact that the 460 and 5850 can overclock bares no relevance. Regarding ROPs and memory of the 1GB 460, from benchmarks I have seen, they make very little difference.

 

[95thrifles]

6 months ago it made sense to recomend the 1gig card as there was v little difference in price, however since the price drop it now makes sense to recomend the 768mb card because as Wayne says paying a 25% premium for a 5-10% performance gain is not value for money

 

[TheRealDeal]

lol, we're talking about saving money and overlocking to get performance past the 460 and even 5850, the fact that the 460 and 5850 can overclock bares no relevance. Regarding ROPs and memory of the 1GB 460, from benchmarks I have has they make very little difference.

Well your agruement about the gtx480 being a better experience than the 5870 is null and void also as if you look at the numbers the gtx480 is around 15% faster than the 5870 but you say its a smoother experience. The same arguement can probably be used here i.e. the gtx480 has more ram than the 5870 same goes for gtx460 1gb over the 768. The gtx460 1gb has around a 10% performance difference not to far off the 15% that the gtx480 holds over the 5870.

My point being numbers only tell you half the truth as you found out going to a gtx480. The same could be said for the gtx460 1gb v 768mb. I wonder if there is anybody on here thats had experience with both of these cards.

 

[Big.Wayne]

Hey arknor!

How novel, for once we don't share the same viewpoint on an item of hardware!

Right lets be having ya, good clean debate, no punches below the belt etc!

they are telling you max , min average ?

Indeed, it seems to have been the way graphics card have been benchmarked since year dot? . . . are you suggesting that conventional benchmark performance data that show you min/max & average is somehow a flawed method?

not frame skip ,stutter , frame skip ,stutter , frame skip anytime the card reachs more than 768 memory

"frame skip ,stutter , frame skip ,stutter" huh?

I'm fully aware of how vRam works and what happens once the limit is exceeded, the GPU has to swap out textures to the main system memory which if it happens often enough can produce the odd micro-pause . . . thing is though most modern GPU's include a form of texture compression technology that keeps the graphic data small enough to fit in local vRam and avoid this scenario you speak of . . .

I would say your using some scaremongering tactics here to give Johnny +2 the "justification" he needs to keep spending more money than he actually needs to? . . . keeping a game running smoothly with lots of Eye-Candy enabled is not just about vRam, its about graphic rendering power, shaders, frequencies etc etc

I've played so many games @ 1920x1200 with a 512MB card that ran perfectly smooth and had all the graphic data stored in local vRam . . . having 768MB of vRam really gives a nice bit of overhead for most people that can enjoy a game without having 16x AA enabled and things like that which are impossible to notice in RealWorld gameplay even when examining a still screenshot side-by-side using a magnifying glass!

as i said min/max are useless when your stuttering along out of memory

This is a myth I think . . . if you can produce some tangible "Facts" to backup this I would be most interested . . . I would imagine you could do some testing with a FPS tracker that showed the framerate across a timeline . . . these "stutters" you speak off would be very easy to note as the framerate counter would most likely dive down to 0FPS for a second?

Can you produce this data? . . .

most games already use more than 768mb of memory at 1080.

Can you please be more explict . . . can you define what you mean by "most games"? . . . and also exactly what graphics settings please beyond the basic screen res? . . . . can you produce any data which shows the games you speak off using more than 768MB of vRam please?

 

[Big.Wayne]

Hello robgmun

there's one big thing you are forgetting. Future proofing, as games use more VRAM the 768 will begin to struggle more and more and the 1GB will pull further ahead and the investment will be clear, 12 months from now the difference in fps between the two will be greater as the 768 will drop to single figures.

"one big thing I am forgetting"

If you would scroll back to post #35 you may find I went into quite a bit of detail about why I think this "Future-Proof" ideology in most cases is flawed . . .

To save you scrolling back I will copy & paste the text again here and colour it yellow so you can't miss it . . . .

as games use more VRAM the 768 will begin to struggle more and more and the 1GB will pull further ahead

This is just another "literal" truth which is no better than advocating

• Someone buy a HexCore because as games get more processor intensive the six core processor will pull ahead
• Someone buy 8GB(2x4GB) of system memory because as games get more memory intensive the 8GB machine will pull ahead

Some people always seem to use this "Future-Proof" ideology as a form of justification to convince themselves (and anyone that doesn't know better) that the extra money is a worthwhile investment . . . but as anyone who has been in the hardware game a while knows . . . at that point in the future when the hardware you purchased today no longer quites cuts the mustard there will always be newer, better, faster technology for sale that totally outmodes your supposed "Future-Proofed" investment . . . all the money ones saves today is money that can be put towards this shiny new tech in the future!

The only person who really needs to "Bulk-up" up technology today is the person who is taking a spaceship ride to mars and will have a problem getting any new technology delivered from the OcUK store!

It's called investment

Buying the item of hardware you "need" is called investment . . . beyond a certain level it's called diminishing returns i.e paying extra £££ for less and less gain today . . .

What you are thinking is called "investment" is actually called "commodity speculation" . . . if you have a time machine and have travelled into the future and done some testing then you are an authority, while your there look at the GPU's for sale in the future and check out their prices, I think you will find that most people in the present day will get better value by paying for the technology they can put to good use "today" and the near future and save their cash £££ for a time in the future when the newer technology arrives to take care of ones future technological "needs" . . .

If you find yourself wanting to spend an ever increasing amount of money on technology that doesn't actually benefit you "Today" in the hear and now your doing something wrong . . . in 2012 the £40-£50 quid saved today by buying a 786MB card can be put towards your next purchase, the 1024MB card owner will not be getting a worthwhile return from his extra expenditure the same as the HD 3850 1GB owner who thought vRam was king and the same as the Geforce 8800GT 1GB owner who thought vRam was king!

 

[robgmun]

i will bring this thread back to life in 12 months and we will see how my HAWK still stands up to modern games compares to someones 460 768 card. That person will be thinking about another card and cough up more where i continue being happy, my 8800GTX was also an investment and that lasted me 4 years and at least 2 generations! In the end if you go cheap you pay twice (or maybe three times)

 

[arknor]

Can you please be more explict . . . can you define what you mean by "most games"? . . . and also exactly what graphics settings please beyond the basic screen res? . . . . can you produce any data which shows the games you speak off using more than 768MB of vRam please?

i guess you misssed this thread posted the other day http://forums.overclockers.co.uk/showthread.php?t=18194652

which has links to a foreign article on video card memory usage

they test various resolutions , game settings for the majority are maxx settings with 4xAA
out of 28 games tested at 1680х1050
1 game used between 1100mb - 1199mb of memory
4 used between 800mb - 899mb
5 used between 700 - 799

this is only at 1680x1050 and this is a wide range of games

at 1920 x1200
10 games are using over 768mb of memory out of 28 games tested.
he full extensive article is here http://translate.google.com/transla...videopamyati_nuzhno_dlya_sovremennyh_igr.html

its clearly obvious even now 768mb of ram on a video card is barely cutting it at settings most people who buy one of these cards are likey to want

This is a myth I think . . . if you can produce some tangible "Facts" to backup this I would be most interested . . . I would imagine you could do some testing with a FPS tracker that showed the framerate across a timeline . . . these "stutters" you speak off would be very easy to note as the framerate counter would most likely dive down to 0FPS for a second?

Can you produce this data? . . .

if you can provide me with a GTX 460 768mb card sure . . .

 

[jakesnake]

@big.waynes benchmarks eairler in the post

these benchmarks are probably taken while playing single player mode while shooting 3 or 4 computer bots for a couple of minutes.
i dont think they reflect online multiplayer with a packed 32man server and how intensive that can be. this is wer the 1gb will shine.

im getting 2 460s in a couple of weeks. im still getting the 1Gb version

 

[arknor]

these benchmarks are probably taken while playing single player mode while shooting 3 or 4 computer bots for a couple of minutes.
i dont think they reflect online multiplayer with a packed 32man server and how intensive that can be. this is wer the 1gb will shine.

im getting 2 460s in a couple of weeks. i think ill get the 1Gb still

i think that would be a wise choice if you are planning to play modern games.

maybe if you dont want to play at max settings or you didnt want AA anyway you could just get some 768MB cards and save yourself some money

 

[jakesnake]

maybe if you dont want to play at max settings or you didnt want AA anyway you could just get some 768MB cards and save yourself some money

yea i could lower my res and disable AA and save a few pounds, but i dont want to 1gb powa lol

 

[FoxEye]

i guess you misssed this thread posted the other day http://forums.overclockers.co.uk/showthread.php?t=18194652

which has links to a foreign article on video card memory usage

they test various resolutions , game settings for the majority are maxx settings with 4xAA
out of 28 games tested at 1680х1050
1 game used between 1100mb - 1199mb of memory
4 used between 800mb - 899mb
5 used between 700 - 799

this is only at 1680x1050 and this is a wide range of games

at 1920 x1200
10 games are using over 768mb of memory out of 28 games tested.

Doesn't that mean that the anand benchmark is already taking this into account?

I.e., the games tested are already showing (very small) performance loss due to using system RAM? In other words, the performance loss from not having the extra memory is < the 4FPS difference in the charts?

Alternatively, anand managed by chance to only test the games using < 768 vRam?

 

[arc@css]

Hi Big.Wayne.

No, I don't have any other benchies to link to. I'm sure if we found others they would vary, in some cases substantially. Many of us have seen many benchmarks where game frame rates are different from others, this is nothing uncommon.

It also depends on the user, as we all have different expectations with regard to acceptable frame rate, how much eye candy we demand, AA etc. Although I am sure the 768mb version is a great card for the medium - upper end of mid gaming, I would simply rather have a little too much power than not enough. Been there too many times I'm afraid.

I think it also depends on what the user has in terms of requirements, what titles, (many more than just a few from one table / graph). There are many factors to potentially consider than just benchmarks in general, which, I might add (benchmarks) can make sometimes make an over-simplification of the situation. They are a guide, 'tis all.

Some folk will undoubtedly be happy with their purchase of a 768 mb card, some that couldn't afford to buy something better 'might' be a little less so. Swings and roundabouts I guess.

 

[95thrifles]

courtesy of techreport, couple months old now so prices will of fluctuated but it gives a good guide for comparitive performance/cost ratios

 

[arknor]

Doesn't that mean that the anand benchmark is already taking this into account?

I.e., the games tested are already showing (very small) performance loss due to using system RAM? In other words, the performance loss from not having the extra memory is < the 4FPS difference in the charts?

Alternatively, anand managed by chance to only test the games using < 768 vRam?

are you dense?

how hard is it to understand game benchmarks tell us 2-3 things MINIMUM , AVERAGE , MAXIMUM.
which tells us 3 simple things , it does not tell you how the game actually ran, it doesnt tell you if its actually smooth framerates.

buy your 768mb card for all i care , you know games already exist that use more than that , dont come crying on the forums if your games judder along at what should be a nice smooth fps